Xin Peng

Xueying Du, Mingwei Liu, Kaixin Wang et al.

In this work, we make the first attempt to evaluate LLMs in a more challenging code generation scenario, i.e. class-level code generation. We first manually construct the first class-level code generation benchmark ClassEval of 100 class-level Python code generation tasks with approximately 500 person-hours. Based on it, we then perform the first study of 11 state-of-the-art LLMs on class-level code generation. Based on our results, we have the following main findings. First, we find that all existing LLMs show much worse performance on class-level code generation compared to on standalone method-level code generation benchmarks like HumanEval; and the method-level coding ability cannot equivalently reflect the class-level coding ability among LLMs. Second, we find that GPT-4 and GPT-3.5 still exhibit dominate superior than other LLMs on class-level code generation, and the second-tier models includes Instruct-Starcoder, Instruct-Codegen, and Wizardcoder with very similar performance. Third, we find that generating the entire class all at once (i.e. holistic generation strategy) is the best generation strategy only for GPT-4 and GPT-3.5, while method-by-method generation (i.e. incremental and compositional) is better strategies for the other models with limited ability of understanding long instructions and utilizing the middle information. Lastly, we find the limited model ability of generating method-dependent code and discuss the frequent error types in generated classes. Our benchmark is available at https://github.com/FudanSELab/ClassEval.

Zhiqiang Yuan, Junwei Liu, Qiancheng Zi et al.

In this work, we evaluate 10 open-source instructed LLMs on four representative code comprehension and generation tasks. We have the following main findings. First, for the zero-shot setting, instructed LLMs are very competitive on code comprehension and generation tasks and sometimes even better than small SOTA models specifically fine-tuned on each downstream task. We also find that larger instructed LLMs are not always better on code-related tasks. Second, for the few-shot setting, we find that adding demonstration examples substantially helps instructed LLMs perform better on most code comprehension and generation tasks; however, the examples would sometimes induce unstable or even worse performance. Furthermore, we find widely-used BM25-based shot selection strategy significantly outperforms the basic random selection or fixed selection only on generation problems. Third, for the fine-tuning setting, we find that fine-tuning could further improve the model performance on downstream code comprehension and generation tasks compared to the zero-shot/one-shot performance. In addition, after being fine-tuned on the same downstream task dataset, instructed LLMs outperform both the small SOTA models and similar-scaled LLMs without instruction tuning. Based on our findings, we further present practical implications on model and usage recommendation, performance and cost trade-offs, and future direction.

Junwei Liu, Kaixin Wang, Yixuan Chen et al.

The recent advance in Large Language Models (LLMs) has shaped a new paradigm of AI agents, i.e., LLM-based agents. Compared to standalone LLMs, LLM-based agents substantially extend the versatility and expertise of LLMs by enhancing LLMs with the capabilities of perceiving and utilizing external resources and tools. To date, LLM-based agents have been applied and shown remarkable effectiveness in Software Engineering (SE). The synergy between multiple agents and human interaction brings further promise in tackling complex real-world SE problems. In this work, we present a comprehensive and systematic survey on LLM-based agents for SE. We collect 124 papers and categorize them from two perspectives, i.e., the SE and agent perspectives. In addition, we discuss open challenges and future directions in this critical domain. The repository of this survey is at https://github.com/FudanSELab/Agent4SE-Paper-List.

Fulin Gao, Weimin Zhong, Zhixing Cao et al.

A desirable open world recognition (OWR) system requires performing three tasks: (1) Open set recognition (OSR), i.e., classifying the known (classes seen during training) and rejecting the unknown (unseen$/$novel classes) online; (2) Grouping and labeling these unknown as novel known classes; (3) Incremental learning (IL), i.e., continual learning these novel classes and retaining the memory of old classes. Ideally, all of these steps should be automated. However, existing methods mostly assume that the second task is completely done manually. To bridge this gap, we propose OpenGCD that combines three key ideas to solve the above problems sequentially: (a) We score the origin of instances (unknown or specifically known) based on the uncertainty of the classifier's prediction; (b) For the first time, we introduce generalized category discovery (GCD) techniques in OWR to assist humans in grouping unlabeled data; (c) For the smooth execution of IL and GCD, we retain an equal number of informative exemplars for each class with diversity as the goal. Moreover, we present a new performance evaluation metric for GCD called harmonic clustering accuracy. Experiments on two standard classification benchmarks and a challenging dataset demonstrate that OpenGCD not only offers excellent compatibility but also substantially outperforms other baselines. Code: https://github.com/Fulin-Gao/OpenGCD.

Wenxuan Tu, Qing Liao, Sihang Zhou et al.

Masked graph autoencoder (MGAE) has emerged as a promising self-supervised graph pre-training (SGP) paradigm due to its simplicity and effectiveness. However, existing efforts perform the mask-then-reconstruct operation in the raw data space as is done in computer vision (CV) and natural language processing (NLP) areas, while neglecting the important non-Euclidean property of graph data. As a result, the highly unstable local connection structures largely increase the uncertainty in inferring masked data and decrease the reliability of the exploited self-supervision signals, leading to inferior representations for downstream evaluations. To address this issue, we propose a novel SGP method termed Robust mAsked gRaph autoEncoder (RARE) to improve the certainty in inferring masked data and the reliability of the self-supervision mechanism by further masking and reconstructing node samples in the high-order latent feature space. Through both theoretical and empirical analyses, we have discovered that performing a joint mask-then-reconstruct strategy in both latent feature and raw data spaces could yield improved stability and performance. To this end, we elaborately design a masked latent feature completion scheme, which predicts latent features of masked nodes under the guidance of high-order sample correlations that are hard to be observed from the raw data perspective. Specifically, we first adopt a latent feature predictor to predict the masked latent features from the visible ones. Next, we encode the raw data of masked samples with a momentum graph encoder and subsequently employ the resulting representations to improve predicted results through latent feature matching. Extensive experiments on seventeen datasets have demonstrated the effectiveness and robustness of RARE against state-of-the-art (SOTA) competitors across three downstream tasks.

Xin Peng, Ling Gao, Yifu Wang et al.

Event cameras are bio-inspired sensors that perform well in challenging illumination conditions and have high temporal resolution. However, their concept is fundamentally different from traditional frame-based cameras. The pixels of an event camera operate independently and asynchronously. They measure changes of the logarithmic brightness and return them in the highly discretised form of time-stamped events indicating a relative change of a certain quantity since the last event. New models and algorithms are needed to process this kind of measurements. The present work looks at several motion estimation problems with event cameras. The flow of the events is modelled by a general homographic warping in a space-time volume, and the objective is formulated as a maximisation of contrast within the image of warped events. Our core contribution consists of deriving globally optimal solutions to these generally non-convex problems, which removes the dependency on a good initial guess plaguing existing methods. Our methods rely on branch-and-bound optimisation and employ novel and efficient, recursive upper and lower bounds derived for six different contrast estimation functions. The practical validity of our approach is demonstrated by a successful application to three different event camera motion estimation problems.

Kaifeng He, Xiaojun Zhang, Peiliang Cai et al.

Large language models (LLMs) frequently generate defective outputs in code generation tasks, ranging from logical bugs to security vulnerabilities. While these generation failures are often treated as model-level limitations, empirical evidence increasingly traces their root causes to imperfections within the training corpora. Yet, the specific mechanisms linking training data quality issues to generated code quality issues remain largely unmapped. This paper presents a systematic literature review of 114 primary studies to investigate how training data quality issues propagate into code generation. We establish a unified taxonomy that categorizes generated code quality issues across nine dimensions and training data quality issues into code and non-code attributes. Based on this taxonomy, we formalize a causal framework detailing 18 typical propagation mapping mechanisms. Furthermore, we synthesize state-of-the-art detection and mitigation techniques across the data, model, and generation lifecycles. The reviewed literature reveals a clear methodological shift: quality assurance is transitioning from reactive, heuristic-based post-generation filtering toward proactive, data-centric governance and closed-loop repair. Finally, we identify open challenges and outline research directions for developing reliable LLMs for code through integrated data curation and continuous evaluation. Our repository is available at https://github.com/SYSUSELab/From-Data-to-Code.

Xin Peng, Yifu Wang, Ling Gao et al.

Event cameras are bio-inspired sensors that perform well in HDR conditions and have high temporal resolution. However, different from traditional frame-based cameras, event cameras measure asynchronous pixel-level brightness changes and return them in a highly discretised format, hence new algorithms are needed. The present paper looks at fronto-parallel motion estimation of an event camera. The flow of the events is modeled by a general homographic warping in a space-time volume, and the objective is formulated as a maximisation of contrast within the image of unwarped events. However, in stark contrast to prior art, we derive a globally optimal solution to this generally non-convex problem, and thus remove the dependency on a good initial guess. Our algorithm relies on branch-and-bound optimisation for which we derive novel, recursive upper and lower bounds for six different contrast estimation functions. The practical validity of our approach is supported by a highly successful application to AGV motion estimation with a downward facing event camera, a challenging scenario in which the sensor experiences fronto-parallel motion in front of noisy, fast moving textures.

Ling Gao, Junyan Su, Jiadi Cui et al.

The motion of planar ground vehicles is often non-holonomic, and as a result may be modelled by the 2 DoF Ackermann steering model. We analyse the feasibility of estimating such motion with a downward facing camera that exerts fronto-parallel motion with respect to the ground plane. This turns the motion estimation into a simple image registration problem in which we only have to identify a 2-parameter planar homography. However, one difficulty that arises from this setup is that ground-plane features are indistinctive and thus hard to match between successive views. We encountered this difficulty by introducing the first globally-optimal, correspondence-less solution to plane-based Ackermann motion estimation. The solution relies on the branch-and-bound optimisation technique. Through the low-dimensional parametrisation, a derivation of tight bounds, and an efficient implementation, we demonstrate how this technique is eventually amenable to accurate real-time motion estimation. We prove its property of global optimality and analyse the impact of assuming a locally constant centre of rotation. Our results on real data finally demonstrate a significant advantage over the more traditional, correspondence-based hypothesise-and-test schemes.

Xuan Jiang, Yuhan Tang, Junzhe Cao et al.

Urban air mobility (UAM) has the potential to revolutionize transportation in metropolitan areas, providing a new mode of transportation that could alleviate congestion and improve accessibility. However, the integration of UAM into existing transportation systems is a complex task that requires a thorough understanding of its impact on traffic flow and capacity. In this paper, we conduct a survey to investigate the current state of research on UAM in metropolitan-scale traffic using simulation techniques. We identify key challenges and opportunities for the integration of UAM into urban transportation systems, including impacts on existing traffic patterns and congestion; safety analysis and risk assessment; potential economic and environmental benefits; and the development of shared infrastructure and routes for UAM and ground-based transportation. We also discuss the potential benefits of UAM, such as reduced travel times and improved accessibility for underserved areas. Our survey provides a comprehensive overview of the current state of research on UAM in metropolitan-scale traffic using simulation and highlights key areas for future research and development.

Junming Cao, Bihuan Chen, Longjie Hu et al.

Machine learning (ML) enabled systems are emerging with recent breakthroughs in ML. A model-centric view is widely taken by the literature to focus only on the analysis of ML models. However, only a small body of work takes a system view that looks at how ML components work with the system and how they affect software engineering for MLenabled systems. In this paper, we adopt this system view, and conduct a case study on Rasa 3.0, an industrial dialogue system that has been widely adopted by various companies around the world. Our goal is to characterize the complexity of such a largescale ML-enabled system and to understand the impact of the complexity on testing. Our study reveals practical implications for software engineering for ML-enabled systems.

Zhiqiang Yuan, Weitong Chen, Hanlin Wang et al.

Code translation converts code from one programming language to another while maintaining its original functionality, which is crucial for software migration, system refactoring, and cross-platform development. Traditional rule-based methods rely on manually-written rules, which can be time-consuming and often result in less readable code. To overcome this, learning-based methods have been developed, leveraging parallel data to train models for automated code translation. More recently, the advance of Large Language Models (LLMs) further boosts learning-based code translation. Although promising, LLM-translated program still suffers from diverse quality issues (e.g., syntax errors and semantic errors). In particular, it can be challenging for LLMs to self-debug these errors when simply provided with the corresponding error messages. In this work, we propose a novel LLM-based multi-agent system TRANSAGENT, which enhances LLM-based code translation by fixing the syntax errors and semantic errors with the synergy between four LLM-based agents, including Initial Code Translator, Syntax Error Fixer, Code Aligner, and Semantic Error Fixer. The main insight of TRANSAGENT is to first localize the error code block in the target program based on the execution alignment between the target and source program, which can narrow down the fixing space and thus lower down the fixing difficulties. To evaluate TRANSAGENT, we first construct a new benchmark from recent programming tasks to mitigate the potential data leakage issue. On our benchmark, TRANSAGENT outperforms the latest LLM-based code translation technique UniTrans in both translation effectiveness and efficiency; additionally, our evaluation on different LLMs show the generalization of TRANSAGENT and our ablation study shows the contribution of each agent.

Kaifeng He, Mingwei Liu, Chong Wang et al.

Code generation with large language models (LLMs) is highly sensitive to token selection during decoding, particularly at uncertain decision points that influence program logic. While standard strategies such as greedy decoding treat all tokens uniformly, they overlook code-specific uncertainty patterns, leading to suboptimal performance. This paper presents an empirical study revealing that many generation errors stem from token ranking mistakes at high-uncertainty steps, where the correct token is present but not top-ranked. Motivated by these findings, we propose AdaDec, a lookahead-based uncertainty-guided adaptive decoding framework that integrates a token-level pause-then-rerank mechanism driven by token uncertainty. AdaDec learns model-specific uncertainty thresholds and applies a lookahead-based reranking strategy when uncertainty is high. Experiments on HumanEval+, MBPP+, and DevEval benchmarks show that AdaDec improves Pass@1 accuracy by up to 20.9% in absolute terms over greedy decoding. More importantly, it consistently outperforms both competitive baselines like Beam Search and state-of-the-art adaptive decoding methods such as AdapT, while maintaining high efficiency through selective, uncertainty-triggered pausing. Our results highlight the promise of uncertainty-aware adaptive decoding for improving both the reliability and efficiency of LLM-based code generation.

Haowen Jiang, Xinyu Huang, You Lu et al.

Recent advancements in end-to-end autonomous driving systems (ADSs) underscore their potential for perception and planning capabilities. However, challenges remain. Complex driving scenarios contain rich semantic information, yet ambiguous or noisy semantics can compromise decision reliability, while interference between multiple driving tasks may hinder optimal planning. Furthermore, prolonged inference latency slows decision-making, increasing the risk of unsafe driving behaviors. To address these challenges, we propose ExpertAD, a novel framework that enhances the performance of ADS with Mixture of Experts (MoE) architecture. We introduce a Perception Adapter (PA) to amplify task-critical features, ensuring contextually relevant scene understanding, and a Mixture of Sparse Experts (MoSE) to minimize task interference during prediction, allowing for effective and efficient planning. Our experiments show that ExpertAD reduces average collision rates by up to 20% and inference latency by 25% compared to prior methods. We further evaluate its multi-skill planning capabilities in rare scenarios (e.g., accidents, yielding to emergency vehicles) and demonstrate strong generalization to unseen urban environments. Additionally, we present a case study that illustrates its decision-making process in complex driving scenarios.

Dingji Wang, You Lu, Bihuan Chen et al.

End-to-end autonomous driving systems (ADSs), with their strong capabilities in environmental perception and generalizable driving decisions, are attracting growing attention from both academia and industry. However, once deployed on public roads, ADSs are inevitably exposed to diverse driving hazards that may compromise safety and degrade system performance. This raises a strong demand for resilience of ADSs, particularly the capability to continuously monitor driving hazards and adaptively respond to potential safety violations, which is crucial for maintaining robust driving behaviors in complex driving scenarios. To bridge this gap, we propose a runtime resilience-oriented framework, Argus, to mitigate the driving hazards, thus preventing potential safety violations and improving the driving performance of an ADS. Argus continuously monitors the trajectories generated by the ADS for potential hazards and, whenever the EGO vehicle is deemed unsafe, seamlessly takes control through a hazard mitigator. We integrate Argus with three state-of-the-art end-to-end ADSs, i.e., TCP, UniAD and VAD. Our evaluation has demonstrated that Argus effectively and efficiently enhances the resilience of ADSs, improving the driving score of the ADS by up to 150.30% on average, and preventing up to 64.38% of the violations, with little additional time overhead.

Xin Peng, Bo Lin, Jing Wang et al.

Automated vulnerability detection is crucial for enhancing software security by identifying potential flaws that attackers could exploit, thereby reducing the reliance on labor-intensive manual code audits. Recent advancements have shifted towards leveraging large language models (LLMs) for vulnerability detection, with techniques like Vul-RAG and VulnSage demonstrating progress through structured prompting and external knowledge integration. However, these approaches typically rely on a single reasoning paradigm, limiting their ability to address the complex and diverse nature of real-world vulnerabilities. To overcome these limitations, we propose ReasonVul, a novel multi-perspective reasoning framework that harnesses cognitive synergy among three specialized LLM agents, each embodying a distinct reasoning mode. The framework begins with independent analyses of the source code, followed by a structured debate mechanism to resolve conflicts through iterative rebuttal and revision, ultimately converging on a collaborative judgment. Evaluated on the PrimeVul dataset, ReasonVul achieves a PairAcc of 40.00% and an F1-score of 72.52%, surpassing the best baseline by 81.24% in PairAcc. Further tests on the JITVUL dataset confirm its generalizability, with a PairAcc of 28.67%. Additionally, we analyzed 542 conflict cases and found that 389 were correctly resolved, highlighting the framework's ability to uncover hidden vulnerabilities through the error-correction mechanism driven by the debate. This work emphasizes the importance of multi-perspective reasoning and collaborative validation in achieving robust and comprehensive vulnerability detection in real-world software systems.

Yin Luo, Qingchao Kong, Nan Xu et al.

As the latest advancements in natural language processing, large language models (LLMs) have achieved human-level language understanding and generation abilities in many real-world tasks, and even have been regarded as a potential path to the artificial general intelligence. To better facilitate research on LLMs, many open-source LLMs, such as Llama 2 and Falcon, have recently been proposed and gained comparable performances to proprietary models. However, these models are primarily designed for English scenarios and exhibit poor performances in Chinese contexts. In this technical report, we propose YAYI 2, including both base and chat models, with 30 billion parameters. YAYI 2 is pre-trained from scratch on a multilingual corpus which contains 2.65 trillion tokens filtered by our pre-training data processing pipeline. The base model is aligned with human values through supervised fine-tuning with millions of instructions and reinforcement learning from human feedback. Extensive experiments on multiple benchmarks, such as MMLU and CMMLU, consistently demonstrate that the proposed YAYI 2 outperforms other similar sized open-source models.

Junwei Liu, Chen Xu, Chong Wang et al.

Recent advances in large language model agents offer the promise of automating end-to-end software development from natural language requirements. However, existing approaches largely adopt linear, waterfall-style pipelines, which oversimplify the iterative nature of real-world development and struggle with complex, large-scale projects. To address these limitations, we propose EvoDev, an iterative software development framework inspired by feature-driven development. EvoDev decomposes user requirements into a set of user-valued features and constructs a Feature Map, a directed acyclic graph that explicitly models dependencies between features. Each node in the feature map maintains multi-level information, including business logic, design, and code, which is propagated along dependencies to provide context for subsequent development iterations. We evaluate EvoDev on challenging Android development tasks and show that it outperforms the best-performing baseline, Claude Code, by a substantial margin of 56.8%, while improving single-agent performance by 16.0%-76.6% across different base LLMs, highlighting the importance of dependency modeling, context propagation, and workflow-aware agent design for complex software projects. Our work summarizes practical insights for designing iterative, LLM-driven development frameworks and informs future training of base LLMs to better support iterative software development.

Yiheng Cao, Yihao Chen, Xin Hu et al.

Detecting vulnerabilities in source code remains critical yet challenging, as conventional static analysis tools construct inaccurate program representations, while existing LLM-based approaches often miss essential vulnerability context and lack grounded reasoning. To mitigate these challenges, we introduce VulWeaver, a novel LLM-based approach that weaves broken program semantics into accurate representations and extracts holistic vulnerability context for grounded vulnerability detection. Specifically, VulWeaver first constructs an enhanced unified dependency graph (UDG) by integrating deterministic rules with LLM-based semantic inference to address static analysis inaccuracies. It then extracts holistic vulnerability context by combining explicit contexts from program slicing with implicit contexts, including usage, definition, and declaration information. Finally, VulWeaver employs meta-prompting with vulnerability type specific expert guidelines to steer LLMs through systematic reasoning, aggregated via majority voting for robustness. Extensive experiments on PrimeVul4J dataset have demonstrated that VulWeaver achieves a F1-score of 0.75, outperforming state-of-the-art learning-based, LLM-based, and agent-based baselines by 23%, 15%, and 60% in F1-score, respectively. VulWeaver has also detected 26 true vulnerabilities across 9 realworld Java projects, with 15 confirmed by developers and 5 CVE identifiers assigned. In industrial deployment, VulWeaver identified 40 confirmed vulnerabilities in an internal repository.

Yiheng Huang, Zhijia Zhao, Bihuan Chen et al.

The model context protocol (MCP) standardizes how LLMs connect to external tools and data sources, enabling faster integration but introducing new attack vectors. Despite the growing adoption of MCP, existing MCP security studies classify attacks by their observable effects, obscuring how attacks behave across different MCP server components and overlooking multi-component attack chains. Meanwhile, existing defenses are less effective when facing multi-component attacks or previously unknown malicious behaviors. This work presents a component-centric perspective for understanding and detecting malicious MCP servers. First, we build the first component-centric PoC dataset of 114 malicious MCP servers where attacks are achieved as manipulation over MCP components and their compositions. We evaluate these attacks' effectiveness across two MCP hosts and five LLMs, and uncover that (1) component position shapes attack success rate; and (2) multi-component compositions often outperform single-component attacks by distributing malicious logic. Second, we propose and implement Connor, a two-stage behavioral deviation detector for malicious MCP servers. It first performs pre-execution analysis to detect malicious shell commands and extract each tool's function intent, and then conducts step-wise in-execution analysis to trace each tool's behavioral trajectories and detect deviations from its function intent. Evaluation on our curated dataset indicates that Connor achieves an F1-score of 94.6%, outperforming the state of the art by 8.9% to 59.6%. In real-world detection, Connor identifies two malicious servers.

Wanting Xu, Si'ao Zhang, Li Cui et al.

Despite the promise of superior performance under challenging conditions, event-based motion estimation remains a hard problem owing to the difficulty of extracting and tracking stable features from event streams. In order to robustify the estimation, it is generally believed that fusion with other sensors is a requirement. In this work, we demonstrate reliable, purely event-based visual odometry on planar ground vehicles by employing the constrained non-holonomic motion model of Ackermann steering platforms. We extend single feature n-linearities for regular frame-based cameras to the case of quasi time-continuous event-tracks, and achieve a polynomial form via variable degree Taylor expansions. Robust averaging over multiple event tracks is simply achieved via histogram voting. As demonstrated on both simulated and real data, our algorithm achieves accurate and robust estimates of the vehicle's instantaneous rotational velocity, and thus results that are comparable to the delta rotations obtained by frame-based sensors under normal conditions. We furthermore significantly outperform the more traditional alternatives in challenging illumination scenarios. The code is available at \url{https://github.com/gowanting/NHEVO}.

Linxi Liang, Jing Gong, Mingwei Liu et al.

Large Language Models (LLMs) have become pivotal tools for automating code generation in software development. However, these models face significant challenges in producing version-aware code for rapidly evolving languages like Rust, where frequent Application Programming Interfaces (API) changes across versions lead to compatibility issues and correctness errors. Existing benchmarks lack systematic evaluation of how models navigate API transitions, relying on labor-intensive manual curation and offering limited version-specific insights. To address this gap, we present RustEvo, a novel framework for constructing dynamic benchmarks that evaluate the ability of LLMs to adapt to evolving Rust APIs. RustEvo automates dataset creation by synthesizing 588 API changes (380 from Rust standard libraries, 208 from 15 third-party crates) into programming tasks mirroring real-world challenges. These tasks cover four API evolution categories: Stabilizations, Signature Changes, Behavioral Changes, and Deprecations, reflecting their actual distribution in the Rust ecosystem. Experiments on state-of-the-art (SOTA) LLMs reveal significant performance variations: models achieve a 65.8% average success rate on stabilized APIs but only 38.0% on behavioral changes, highlighting difficulties in detecting semantic shifts without signature alterations. Knowledge cutoff dates strongly influence performance, with models scoring 56.1% on before-cutoff APIs versus 32.5% on after-cutoff tasks. Retrieval-Augmented Generation (RAG) mitigates this gap, improving success rates by 13.5% on average for APIs released after model training. Our findings underscore the necessity of our evolution-aware benchmarks to advance the adaptability of LLMs in fast-paced software ecosystems. The framework and the benchmarks are publicly released at https://github.com/SYSUSELab/RustEvo.

Jinze Yu, Xin Peng, Zhengda Lu et al.

A spike camera is a specialized high-speed visual sensor that offers advantages such as high temporal resolution and high dynamic range compared to conventional frame cameras. These features provide the camera with significant advantages in many computer vision tasks. However, the tasks of novel view synthesis based on spike cameras remain underdeveloped. Although there are existing methods for learning neural radiance fields from spike stream, they either lack robustness in extremely noisy, low-quality lighting conditions or suffer from high computational complexity due to the deep fully connected neural networks and ray marching rendering strategies used in neural radiance fields, making it difficult to recover fine texture details. In contrast, the latest advancements in 3DGS have achieved high-quality real-time rendering by optimizing the point cloud representation into Gaussian ellipsoids. Building on this, we introduce SpikeGS, the method to learn 3D Gaussian fields solely from spike stream. We designed a differentiable spike stream rendering framework based on 3DGS, incorporating noise embedding and spiking neurons. By leveraging the multi-view consistency of 3DGS and the tile-based multi-threaded parallel rendering mechanism, we achieved high-quality real-time rendering results. Additionally, we introduced a spike rendering loss function that generalizes under varying illumination conditions. Our method can reconstruct view synthesis results with fine texture details from a continuous spike stream captured by a moving spike camera, while demonstrating high robustness in extremely noisy low-light scenarios. Experimental results on both real and synthetic datasets demonstrate that our method surpasses existing approaches in terms of rendering quality and speed.

Mingwei Liu, Juntao Li, Ying Wang et al.

Despite recent advances in Large Language Models (LLMs) for code generation, the quality of LLM-generated code still faces significant challenges. One significant issue is code repetition, which refers to the model's tendency to generate structurally redundant code, resulting in inefficiencies and reduced readability. To address this, we conduct the first empirical study to investigate the prevalence and nature of repetition across 19 state-of-the-art code LLMs using three widely-used benchmarks. Our study includes both quantitative and qualitative analyses, revealing that repetition is pervasive and manifests at various granularities and extents, including character, statement, and block levels. We further summarize a taxonomy of 20 repetition patterns. Building on our findings, we propose DeRep, a rule-based technique designed to detect and mitigate repetition in generated code. We evaluate DeRep using both open-source benchmarks and in an industrial setting. Our results demonstrate that DeRep significantly outperforms baselines in reducing repetition (with an average improvements of 91.3%, 93.5%, and 79.9% in rep-3, rep-line, and sim-line metrics) and enhancing code quality (with a Pass@1 increase of 208.3% over greedy search). Furthermore, integrating DeRep improves the performance of existing repetition mitigation methods, with Pass@1 improvements ranging from 53.7% to 215.7%.

Xueying Du, Jiayi Feng, Yi Zou et al.

Static analysis tools (SATs) are widely adopted in both academia and industry for improving software quality, yet their practical use is often hindered by high false positive rates, especially in large-scale enterprise systems. These false alarms demand substantial manual inspection, creating severe inefficiencies in industrial code review. While recent work has demonstrated the potential of large language models (LLMs) for false alarm reduction on open-source benchmarks, their effectiveness in real-world enterprise settings remains unclear. To bridge this gap, we conduct the first comprehensive empirical study of diverse LLM-based false alarm reduction techniques in an industrial context at Tencent, one of the largest IT companies in China. Using data from Tencent's enterprise-customized SAT on its large-scale Advertising and Marketing Services software, we construct a dataset of 433 alarms (328 false positives, 105 true positives) covering three common bug types. Through interviewing developers and analyzing the data, our results highlight the prevalence of false positives, which wastes substantial manual effort (e.g., 10-20 minutes of manual inspection per alarm). Meanwhile, our results show the huge potential of LLMs for reducing false alarms in industrial settings (e.g., hybrid techniques of LLM and static analysis eliminate 94-98% of false positives with high recall). Furthermore, LLM-based techniques are cost-effective, with per-alarm costs as low as 2.1-109.5 seconds and $0.0011-$0.12, representing orders-of-magnitude savings compared to manual review. Finally, our case analysis further identifies key limitations of LLM-based false alarm reduction in industrial settings.

Dayu Tan, Ziwei Zhang, Yansan Su et al.

Numerous CNN-Transformer hybrid models rely on high-complexity global attention mechanisms to capture long-range dependencies, which introduces non-linear computational complexity and leads to significant resource consumption. Although knowledge distillation and sparse attention mechanisms can improve efficiency, they often fall short of delivering the high segmentation accuracy necessary for complex tasks. Balancing model performance with computational efficiency remains a critical challenge. In this work, we propose a novel 3D multi-modal image segmentation framework, termed MSD-KMamba, which integrates bidirectional spatial perception with multi-scale self-distillation. The bidirectional spatial aware branch effectively captures long-range spatial context dependencies across brain regions, while also incorporating a powerful nonlinear feature extraction mechanism that further enhances the model's ability to learn complex and heterogeneous patterns. In addition, the proposed multi-scale self-distilled fusion strategy strengthens hierarchical feature representations and improves the transfer of semantic information at different resolution levels. By jointly leveraging the bidirectional spatial perception branch and the multi-scale self-distilled fusion strategy, our framework effectively mitigates the bottleneck of quadratic computational complexity in volumetric segmentation, while simultaneously addressing the limitation of insufficient global perception. Extensive experiments on multiple standard benchmark datasets demonstrate that MSD-KMamba consistently outperforms state-of-the-art methods in segmentation accuracy, robustness, and generalization, while maintaining high computational efficiency and favorable scalability. The source code of MSD-KMamba is publicly available at https://github.com/daimao-zhang/MSD-KMamba.

Dayu Tan, Zhenpeng Xu, Yansen Su et al.

Both local details and global context are crucial in medical image segmentation, and effectively integrating them is essential for achieving high accuracy. However, existing mainstream methods based on CNN-Transformer hybrid architectures typically employ simple feature fusion techniques such as serial stacking, endpoint concatenation, or pointwise addition, which struggle to address the inconsistencies between features and are prone to information conflict and loss. To address the aforementioned challenges, we innovatively propose HiPerformer. The encoder of HiPerformer employs a novel modular hierarchical architecture that dynamically fuses multi-source features in parallel, enabling layer-wise deep integration of heterogeneous information. The modular hierarchical design not only retains the independent modeling capability of each branch in the encoder, but also ensures sufficient information transfer between layers, effectively avoiding the degradation of features and information loss that come with traditional stacking methods. Furthermore, we design a Local-Global Feature Fusion (LGFF) module to achieve precise and efficient integration of local details and global semantic information, effectively alleviating the feature inconsistency problem and resulting in a more comprehensive feature representation. To further enhance multi-scale feature representation capabilities and suppress noise interference, we also propose a Progressive Pyramid Aggregation (PPA) module to replace traditional skip connections. Experiments on eleven public datasets demonstrate that the proposed method outperforms existing segmentation techniques, demonstrating higher segmentation accuracy and robustness. The code is available at https://github.com/xzphappy/HiPerformer.

Zekai Zhang, Mingwei Liu, Zhenxi Chen et al.

Code editing plays a vital role in software engineering, requiring developers to adjust existing code according to natural language instructions while keeping functionality intact and avoiding unnecessary modifications. However, commit-based datasets commonly used for this task are often noisy, lack diversity, and fail to reflect the style of real-world edit instructions. To address this, we introduce OpenCodeEdit, an open-source pipeline that leverages multiple LLMs to synthesize realistic code-edit triplets. The pipeline produces both concise "lazy" instructions and more detailed "descriptive" ones, and applies filtering based on diffs and topics to guarantee data quality and variety. Using this process, we construct OCEDataFT, a curated dataset of 20K samples. Fine-tuning three advanced base models on OCEDataFT leads to significant performance boosts on the CanItEdit benchmark, with relative pass@1 improvements ranging from 4.50% to 20.79%. Notably, the resulting models achieve performance close to closed-source systems, narrowing the gap to GPT-4 to just 3.54%, without relying on proprietary resources or manual annotation.

Zhenhao Zhou, Zhuochen Huang, Yike He et al.

The Linux kernel is a critical system, serving as the foundation for numerous systems. Bugs in the Linux kernel can cause serious consequences, affecting billions of users. Fault localization (FL), which aims at identifying the buggy code elements in software, plays an essential role in software quality assurance. While recent LLM agents have achieved promising accuracy in FL on recent benchmarks like SWE-bench, it remains unclear how well these methods perform in the Linux kernel, where FL is much more challenging due to the large-scale code base, limited observability, and diverse impact factors. In this paper, we introduce LinuxFLBench, a FL benchmark constructed from real-world Linux kernel bugs. We conduct an empirical study to assess the performance of state-of-the-art LLM agents on the Linux kernel. Our initial results reveal that existing agents struggle with this task, achieving a best top-1 accuracy of only 41.6% at file level. To address this challenge, we propose LinuxFL$^+$, an enhancement framework designed to improve FL effectiveness of LLM agents for the Linux kernel. LinuxFL$^+$ substantially improves the FL accuracy of all studied agents (e.g., 7.2% - 11.2% accuracy increase) with minimal costs. Data and code are available at https://github.com/FudanSELab/LinuxFLBench.

Congying Xu, Bihuan Chen, Chenhao Lu et al.

Open source software (OSS) vulnerabilities threaten the security of software systems that use OSS. Vulnerability databases provide valuable information (e.g., vulnerable version and patch) to mitigate OSS vulnerabilities. There arises a growing concern about the information quality of vulnerability databases. However, it is unclear what the quality of patches in existing vulnerability databases is; and existing manual or heuristic-based approaches for patch tracking are either too expensive or too specific to apply to all OSS vulnerabilities.

Chi Chen, Xin Peng, Zhenchang Xing et al.

Context based API recommendation is an important way to help developers find the needed APIs effectively and efficiently. For effective API recommendation, we need not only a joint view of both structural and textual code information, but also a holistic view of correlated API usage in control and data flow graph as a whole. Unfortunately, existing API recommendation methods exploit structural or textual code information separately. In this work, we propose a novel API recommendation approach called APIRec-CST (API Recommendation by Combining Structural and Textual code information). APIRec-CST is a deep learning model that combines the API usage with the text information in the source code based on an API Context Graph Network and a Code Token Network that simultaneously learn structural and textual features for API recommendation. We apply APIRec-CST to train a model for JDK library based on 1,914 open-source Java projects and evaluate the accuracy and MRR (Mean Reciprocal Rank) of API recommendation with another 6 open-source projects. The results show that our approach achieves respectively a top-1, top-5, top-10 accuracy and MRR of 60.3%, 81.5%, 87.7% and 69.4%, and significantly outperforms an existing graph-based statistical approach and a tree-based deep learning approach for API recommendation. A further analysis shows that textual code information makes sense and improves the accuracy and MRR. We also conduct a user study in which two groups of students are asked to finish 6 programming tasks with or without our APIRec-CST plugin. The results show that APIRec-CST can help the students to finish the tasks faster and more accurately and the feedback on the usability is overwhelmingly positive.

Ying Wang, Bihuan Chen, Kaifeng Huang et al.

Third-party libraries are a central building block to develop software systems. However, outdated third-party libraries are commonly used, and developers are usually less aware of the potential risks. Therefore, a quantitative and holistic study on usages, updates and risks of third-party libraries can provide practical insights to improve the ecosystem sustainably. In this paper, we conduct such a study in the Java ecosystem. Specifically, we conduct a library usage analysis (e.g., usage intensity and outdatedness) and a library update analysis (e.g., update intensity and delay) using 806 open-source projects. The two analyses aim to quantify usage and update practices holistically from the perspective of both open-source projects and third-party libraries. Then, we conduct a library risk analysis (e.g., potential risk and developer response) in terms of bugs with 15 popularly-used third-party libraries. This analysis aims to quantify the potential risk of using outdated libraries and the developer response to the risk. Our findings from the three analyses provide practical insights to developers and researchers on problems and potential solutions in maintaining third-party libraries (e.g., smart alerting and automated updating of outdated libraries). To demonstrate the usefulness of our findings, we propose a bug-driven alerting system for assisting developers to make confident decisions in updating third-party library versions. We have released our dataset to foster valuable applications and improve the ecosystem.

Xin Peng, Ang Gao

We introduce a new generative modeling framework, \textbf{Flow Matching with Arbitrary Auxiliary Paths (AuxPath-FM)}, which generalizes conditional flow matching by incorporating an auxiliary variable drawn from an arbitrary distribution into the probability path. Unlike prior methods that restrict auxiliary components to Gaussian noise, AuxPath-FM allows the variable $η$ to follow any distribution, producing trajectories of the form $X_t = a(t)X_1 + b(t)X_0 + c(t)η$. We theoretically demonstrate that this construction preserves the continuity equation and maintains a training objective consistent with the marginal formulation. This flexibility enables the design of diverse probability paths using various priors, including Gaussian, Uniform, Laplace, and discrete Rademacher distributions, each offering unique geometric properties for generative flows. Furthermore, our framework allows for specialized tasks such as label-guided generation by encoding structured semantic information into the auxiliary distribution. Overall, AuxPath-FM provides a principled and general foundation for probability path design, offering both theoretical generality and practical flexibility for diverse generative modeling tasks.

Xin Peng, Ang Gao

Classifier-Free Guidance (CFG) is essential for high-fidelity conditional generation in flow matching, yet it imposes significant computational overhead by requiring dual forward passes at each sampling step. In this work, we address this bottleneck by introducing \textbf{P-Guide}, a framework that achieves high-quality guidance through a single inference pass by modulating only the initial latent state. We further show that, under a first-order approximation, P-Guide is equivalent to CFG in the sense that it steers generation from the prior space, without requiring explicit velocity field extrapolation during sampling. We consider both homoscedastic and \textbf{heteroscedastic} priors, and find that jointly modeling the mean and variance enables adaptive loss attenuation and improved robustness to data uncertainty. Extensive experiments demonstrate that P-Guide reduces inference latency by approximately 50\% while maintaining fidelity and prompt alignment competitive with standard dual-pass CFG baselines.

Xin Peng, Ang Gao

Flow-based generative models have been employed for sampling the Boltzmann distribution, but their application to high-dimensional systems is hindered by the significant computational cost of obtaining the Jacobian of the flow. To overcome this challenge, we introduce the flow perturbation method, which incorporates optimized stochastic perturbations into the flow. By reweighting trajectories generated by the perturbed flow, our method achieves unbiased sampling of the Boltzmann distribution with orders of magnitude speedup compared to both brute force Jacobian calculations and the Hutchinson estimator. Notably, it accurately sampled the Chignolin protein with all atomic Cartesian coordinates explicitly represented, which, to our best knowledge, is the largest molecule ever Boltzmann sampled in such detail using generative models.

Zhi Qiu, Jiazheng Sun, Chenxiao Xia et al.

While Large Language Models demonstrate remarkable proficiency in high-level semantic planning, they remain limited in handling fine-grained, low-level web component manipulations. To address this limitation, extensive research has focused on enhancing model grounding capabilities through techniques such as Reinforcement Learning. However, rather than compelling agents to adapt to human-centric interfaces, we propose constructing interaction interfaces specifically optimized for agents. This paper introduces Component Interface for Agent (CI4A), a semantic encapsulation mechanism that abstracts the complex interaction logic of UI components into a set of unified tool primitives accessible to agents. We implemented CI4A within Ant Design, an industrial-grade front-end framework, covering 23 categories of commonly used UI components. Furthermore, we developed a hybrid agent featuring an action space that dynamically updates according to the page state, enabling flexible invocation of available CI4A tools. Leveraging the CI4A-integrated Ant Design, we refactored and upgraded the WebArena benchmark to evaluate existing SoTA methods. Experimental results demonstrate that the CI4A-based agent significantly outperforms existing approaches, achieving a new SoTA task success rate of 86.3%, alongside substantial improvements in execution efficiency.

Jiazheng Sun, Mingxuan Li, Yingying Zhang et al.

Benchmarks are paramount for gauging progress in the domain of Mobile GUI Agents. In practical scenarios, users frequently fail to articulate precise directives containing full task details at the onset, and their expressions are typically ambiguous. Consequently, agents are required to converge on the user's true intent via active clarification and interaction during execution. However, existing benchmarks predominantly operate under the idealized assumption that user-issued instructions are complete and unequivocal. This paradigm focuses exclusively on assessing single-turn execution while overlooking the alignment capability of the agent. To address this limitation, we introduce AmbiBench, the first benchmark incorporating a taxonomy of instruction clarity to shift evaluation from unidirectional instruction following to bidirectional intent alignment. Grounded in Cognitive Gap theory, we propose a taxonomy of four clarity levels: Detailed, Standard, Incomplete, and Ambiguous. We construct a rigorous dataset of 240 ecologically valid tasks across 25 applications, subject to strict review protocols. Furthermore, targeting evaluation in dynamic environments, we develop MUSE (Mobile User Satisfaction Evaluator), an automated framework utilizing an MLLM-as-a-judge multi-agent architecture. MUSE performs fine-grained auditing across three dimensions: Outcome Effectiveness, Execution Quality, and Interaction Quality. Empirical results on AmbiBench reveal the performance boundaries of SoTA agents across different clarity levels, quantify the gains derived from active interaction, and validate the strong correlation between MUSE and human judgment. This work redefines evaluation standards, laying the foundation for next-generation agents capable of truly understanding user intent.

Mingwei Liu, Zheng Pei, Yanlin Wang et al.

In the context of software frameworks with limited resources (such as HarmonyOS), large language models (LLMs) often exhibit poor code generation performance because they lack sufficient exposure to such environments during pre-training. Although LLMs can usually maintain correct logical structures across programming languages, they frequently struggle when dealing with framework-specific APIs or syntax, resulting in errors. This indicates that while pre-training equips LLMs with general algorithmic capabilities, they remain unfamiliar with the distinctive syntax and API usage of underrepresented frameworks. As a result, even advanced commercial models like GPT-4o cannot reliably generate correct code without prior adaptation. To address this issue, we propose APIKG4SYN, a framework designed to exploit API knowledge graphs for the construction of API-oriented question-code pairs, specifically tailored for low-resource frameworks without requiring executable code. APIKG4SYN integrates both single-API and multi-API knowledge, where the latter is derived through uncertainty estimation (UE)-driven Monte Carlo Tree Search (MCTS), enabling the creation of a diverse and informative dataset for fine-tuning LLMs. Using HarmonyOS as a case study, we build the first benchmark for HarmonyOS code generation. Experimental results show that fine-tuning Qwen with APIKG4SYN raises pass@1 accuracy to 25.00%, compared with 17.59% for the baseline GPT model. These results confirm that API-oriented data significantly enhance LLM performance in low-resource software development scenarios.

Wanting Xu, Xin Peng, Laurent Kneip

Traditional visual-inertial state estimation targets absolute camera poses and spatial landmark locations while first-order kinematics are typically resolved as an implicitly estimated sub-state. However, this poses a risk in velocity-based control scenarios, as the quality of the estimation of kinematics depends on the stability of absolute camera and landmark coordinates estimation. To address this issue, we propose a novel solution to tight visual-inertial fusion directly at the level of first-order kinematics by employing a dynamic vision sensor instead of a normal camera. More specifically, we leverage trifocal tensor geometry to establish an incidence relation that directly depends on events and camera velocity, and demonstrate how velocity estimates in highly dynamic situations can be obtained over short time intervals. Noise and outliers are dealt with using a nested two-layer RANSAC scheme. Additionally, smooth velocity signals are obtained from a tight fusion with pre-integrated inertial signals using a sliding window optimizer. Experiments on both simulated and real data demonstrate that the proposed tight event-inertial fusion leads to continuous and reliable velocity estimation in highly dynamic scenarios independently of absolute coordinates. Furthermore, in extreme cases, it achieves more stable and more accurate estimation of kinematics than traditional, point-position-based visual-inertial odometry.

Xueying Du, Mingwei Liu, Hanlin Wang et al.

Software crash bugs cause unexpected program behaviors or even abrupt termination, thus demanding immediate resolution. However, resolving crash bugs can be challenging due to their complex root causes, which can originate from issues in the source code or external factors like third-party library dependencies. Large language models (LLMs) have shown promise in software engineering tasks. However, existing research predominantly focuses on the capability of LLMs to localize and repair code-related crash bugs, leaving their effectiveness in resolving environment-related crash bugs in real-world software unexplored. To fill this gap, we conducted the first comprehensive study to assess the capability of LLMs in resolving real-world environment-related crash bugs. We first systematically compare LLMs' performance in resolving code-related and environment-related crash bugs with varying levels of crash contextual information. Our findings reveal that localization is the primary challenge for resolving code-related crashes, while repair poses a greater challenge for environment-related crashes. Furthermore, we investigate the impact of different prompt strategies on improving the resolution of environment-related crash bugs, incorporating different prompt templates and multi-round interactions. Building on this, we further explore an advanced active inquiry prompting strategy leveraging the self-planning capabilities of LLMs. Based on these explorations, we propose IntDiagSolver, an interactive methodology designed to enable precise crash bug resolution through ongoing engagement with LLMs. Extensive evaluations of IntDiagSolver across multiple LLMs (including GPT-3.5, GPT-4, Claude, CodeLlama, DeepSeek-R1, and Qwen-3-Coder) demonstrate consistent improvements in resolution accuracy, with substantial enhancements ranging from 9.1% to 43.3% in localization and 9.1% to 53.3% in repair.

Xin Peng, Ang Gao

Generative models, including diffusion and flow-based models, often exhibit systematic biases that degrade sample quality, particularly in high-dimensional settings. We revisit refinement methods and show that effective bias correction can be achieved as a post-hoc procedure, without noise injection or multi-step resampling of the sampling process. We propose a flow-matching-based \textbf{Bi-stage Flow Refinement (BFR)} framework with two refinement strategies operating at different stages: latent space alignment for approximately invertible generators and data space refinement trained with lightweight augmentations. Unlike previous refiners that perturb sampling dynamics, BFR preserves the original ODE trajectory and applies deterministic corrections to generated samples. Experiments on MNIST, CIFAR-10, and FFHQ at 256x256 resolution demonstrate consistent improvements in fidelity and coverage; notably, starting from base samples with FID 3.95, latent space refinement achieves a \textbf{state-of-the-art} FID of \textbf{1.46} on MNIST using only a single additional function evaluation (1-NFE), while maintaining sample diversity.

Xin Peng, Ang Gao

The scalability of continuous normalizing flows (CNFs) for unbiased Boltzmann sampling remains limited in high-dimensional systems due to the cost of Jacobian-determinant evaluation, which requires $D$ backpropagation passes through the flow layers. Existing stochastic Jacobian estimators such as the Hutchinson trace estimator reduce computation but introduce bias, while the recently proposed Flow Perturbation method is unbiased yet suffers from high variance. We present \textbf{Flow Perturbation++}, a variance-reduced extension of Flow Perturbation that discretizes the probability-flow ODE and performs unbiased stepwise Jacobian estimation at each integration step. This multi-step construction retains the unbiasedness of Flow Perturbation while achieves substantially lower estimator variance. Integrated into a Sequential Monte Carlo framework, Flow Perturbation++ achieves significantly improved equilibrium sampling on a 1000D Gaussian Mixture Model and the all-atom Chignolin protein compared with Hutchinson-based and single-step Flow Perturbation baselines.

Tong Su, Xin Peng, Sarubi Thillainathan et al.

Parameter-efficient fine-tuning (PEFT) methods are increasingly vital in adapting large-scale pre-trained language models for diverse tasks, offering a balance between adaptability and computational efficiency. They are important in Low-Resource Language (LRL) Neural Machine Translation (NMT) to enhance translation accuracy with minimal resources. However, their practical effectiveness varies significantly across different languages. We conducted comprehensive empirical experiments with varying LRL domains and sizes to evaluate the performance of 8 PEFT methods with in total of 15 architectures using the SacreBLEU score. We showed that 6 PEFT architectures outperform the baseline for both in-domain and out-domain tests and the Houlsby+Inversion adapter has the best performance overall, proving the effectiveness of PEFT methods.

Kai Yu, Zhenhao Zhou, Junhao Zeng et al.

Repository-level issue resolution benchmarks have become a standard testbed for evaluating LLM-based agents, yet success is still predominantly measured by test pass rates. In practice, however, acceptable patches must also comply with project-specific design constraints, such as architectural conventions, error-handling policies, and maintainability requirements, which are rarely encoded in tests and are often documented only implicitly in code review discussions. This paper introduces \textit{design-aware issue resolution} and presents \bench{}, a benchmark that makes such implicit design constraints explicit and measurable. \bench{} is constructed by mining and validating design constraints from real-world pull requests, linking them to issue instances, and automatically checking patch compliance using an LLM-based verifier, yielding 495 issues and 1,787 validated constraints across six repositories, aligned with SWE-bench-Verified and SWE-bench-Pro. Experiments with state-of-the-art agents show that test-based correctness substantially overestimates patch quality: fewer than half of resolved issues are fully design-satisfying, design violations are widespread, and functional correctness exhibits negligible statistical association with design satisfaction. While providing issue-specific design guidance reduces violations, substantial non-compliance remains, highlighting a fundamental gap in current agent capabilities and motivating design-aware evaluation beyond functional correctness.

Shengbo Wang, Mingwei Liu, Zike Li et al.

The rapid advancement of Large Language Models (LLMs) poses a significant challenge to existing mathematical reasoning benchmarks. However, these benchmarks tend to become easier over time as LLMs can learn from the published benchmarks. This limitation hinder the precise evaluation of the true capabilities of SOTA models. To address this challenge, this paper introduces EvolMathEval, an automated mathematical benchmark generation and evolution framework based on evolutionary testing. Experimental results demonstrate that EvolMathEval can not only generate a large volume of high-difficulty problems through continuous self-iteration, but it can also significantly enhance the complexity of public datasets like GSM8K through evolution, reducing model accuracy by an average of 48\%. Deeper investigation reveals that when solving these evolved problems, LLMs tend to bypass complex multi-step logical reasoning by relying on simplistic and fuzzy conditions, consequently leading to incorrect solutions. We define this phenomenon as the ``Pseudo Aha Moment", which we find accounts for 77\% to 100\% of errors on targeted problems. Code and resources are available at: https://anonymous.4open.science/r/EvolMathEval

Jianghao Lin, Yuanyuan Shi, Xin Peng et al.

Large language models (LLMs) are increasingly demonstrating strong capabilities as autonomous agents, with function calling serving as a core mechanism for interaction with the environment. Meanwhile, inference scaling has become a cutting-edge technique to enhance LLM performance by allocating more computational resources during the inference process. However, current research on inference scaling primarily focuses on unstructured output generation tasks, leaving its application in structured outputs, like function calling, largely underexplored. To bridge this gap, we propose an inference scaling framework that combines fine-grained beam search with a process reward model, ToolPRM, which scores the internal steps of each single function call. To train ToolPRM, we construct the first fine-grained intra-call process supervision dataset, automatically annotated with function-masking techniques to provide step-level rewards for structured tool-use reasoning. Extensive experiments demonstrate that ToolPRM beats the coarse-grained and outcome reward models in terms of predictive accuracy, indicating its stronger capability in supervising the function calling inference process. Inference scaling technique equipped with ToolPRM also significantly improves the backbone model performance across various function calling tasks and benchmarks. More importantly, we reveal a key principle for applying inference scaling techniques to structured outputs: "explore more but retain less" due to the unrecoverability characteristics of structured function calling generation.

Zhiqiang Yuan, Wenjun Mao, Zhuo Chen et al.

Translating C code into safe Rust is an effective way to ensure its memory safety. Compared to rule-based translation which produces Rust code that remains largely unsafe, LLM-based methods can generate more idiomatic and safer Rust code because LLMs have been trained on vast amount of human-written idiomatic code. Although promising, existing LLM-based methods still struggle with project-level C-to-Rust translation. They typically partition a C project into smaller units (\eg{} functions) based on call graphs and translate them bottom-up to resolve program dependencies. However, this bottom-up, unit-by-unit paradigm often fails to translate pointers due to the lack of a global perspective on their usage. To address this problem, we propose a novel C-Rust Pointer Knowledge Graph (KG) that enriches a code-dependency graph with two types of pointer semantics: (i) pointer-usage information which record global behaviors such as points-to flows and map lower-level struct usage to higher-level units; and (ii) Rust-oriented annotations which encode ownership, mutability, nullability, and lifetime. Synthesizing the \kg{} with LLMs, we further propose \ourtool{}, which implements a project-level C-to-Rust translation technique. In \ourtool{}, the \kg{} provides LLMs with comprehensive pointer semantics from a global perspective, thus guiding LLMs towards generating safe and idiomatic Rust code from a given C project. Our experiments show that \ourtool{} reduces unsafe usages in translated Rust by 99.9\% compared to both rule-based translation and traditional LLM-based rewriting, while achieving an average 29.3\% higher functional correctness than those fuzzing-enhanced LLM methods.

Huacan Chai, Zijie Cao, Maolin Ran et al.

Large language models (LLMs) have achieved impressive success in single-turn function calling, yet real-world applications such as travel planning or multi-stage data analysis typically unfold across multi-turn conversations. In these settings, LLMs must not only issue accurate function calls at each step but also maintain progress awareness, the ability to summarize past interactions and plan future actions to ensure coherent, long-horizon task execution. Existing approaches, however, either reduce multi-turn training to isolated single-turn samples, which neglects task-level planning, or employ end-to-end reinforcement learning (RL) that struggles with redundancy and lacks explicit integration of progress awareness. To overcome these limitations, we introduce PARL-MT, a framework that explicitly incorporates progress awareness into LLM training for multi-turn function calling. PARL-MT combines (i) a Progress Awareness Generation (PAG) pipeline, which automatically constructs datasets coupling conversation summaries with future task planning, and (ii) a Progress Awareness-Guided Reinforcement Learning (PAG-RL) algorithm, which integrates progress awareness into RL training to reduce contextual redundancy and improve alignment between local actions and global task completion. Empirical results on two public benchmarks demonstrate that PARL-MT significantly outperforms existing methods, highlighting the effectiveness of progress awareness in enabling robust and efficient multi-turn function calling.

Jiazheng Sun, Te Yang, Jiayang Niu et al.

Large multi-modal models (LMMs) have advanced mobile GUI agents. However, existing methods struggle with real-world scenarios involving diverse app interfaces and evolving user needs. End-to-end methods relying on model's commonsense often fail on long-tail apps, and agents without user interaction act unilaterally, harming user experience. To address these limitations, we propose Fairy, an interactive multi-agent mobile assistant capable of continuously accumulating app knowledge and self-evolving during usage. Fairy enables cross-app collaboration, interactive execution, and continual learning through three core modules:(i) a Global Task Planner that decomposes user tasks into sub-tasks from a cross-app view; (ii) an App-Level Executor that refines sub-tasks into steps and actions based on long- and short-term memory, achieving precise execution and user interaction via four core agents operating in dual loops; and (iii) a Self-Learner that consolidates execution experience into App Map and Tricks. To evaluate Fairy, we introduce RealMobile-Eval, a real-world benchmark with a comprehensive metric suite, and LMM-based agents for automated scoring. Experiments show that Fairy with GPT-4o backbone outperforms the previous SoTA by improving user requirement completion by 33.7% and reducing redundant steps by 58.5%, showing the effectiveness of its interaction and self-learning.

You Lu, Dingji Wang, Kaifeng Huang et al.

Autonomous vehicle technology has been developed in the last decades with recent advances in sensing and computing technology. There is an urgent need to ensure the reliability and robustness of autonomous driving systems (ADSs). Despite the recent achievements in testing various ADS modules, little attention has been paid on the automated testing of traffic light detection models in ADSs. A common practice is to manually collect and label traffic light data. However, it is labor-intensive, and even impossible to collect diverse data under different driving environments. To address these problems, we propose and implement TigAug to automatically augment labeled traffic light images for testing traffic light detection models in ADSs. We construct two families of metamorphic relations and three families of transformations based on a systematic understanding of weather environments, camera properties, and traffic light properties. We use augmented images to detect erroneous behaviors of traffic light detection models by transformation-specific metamorphic relations, and to improve the performance of traffic light detection models by retraining. Large-scale experiments with four state-of-the-art traffic light detection models and two traffic light datasets have demonstrated that i) TigAug is effective in testing traffic light detection models, ii) TigAug is efficient in synthesizing traffic light images, and iii) TigAug generates traffic light images with acceptable naturalness.